The role of coronary revascularization with either coronary artery bypass grafting (CABG) or percutaneous coronary intervention (PCI) in patients with ischemic cardiomyopathy remains poorly defined (1). The current AHA/ACC guidelines classify the use of CABG in patients with left ventricular ejection fraction (LVEF) <35% as a IIb indication and no recommendations are given for PCI in ischemic cardiomyopathy (2). The Surgical Treatment for Ischemic Cardiomyopathy (STICH) trial (3) conducted between July 2002 and May 2007 randomized 1,212 patients with LVEF <35% and by-passable vessels to CABG plus guideline-directed medical therapy (GDMT) versus GDMT alone. At six years, there was no statistically significant difference in all-cause mortality between CABG and GDMT (41% vs. 36%; hazard ratio [HR]: 0.86; 95% confidence interval [CI]: 0.72 to 1.04; p = 0.12) based on intention-to-treat (3). The lack of significance was thought to be due to the delayed benefit in revascularization, only outweighing the post-operative mortality with sufficient follow-up duration. This phenomenon was borne out in the long term follow-up Surgical Treatment for Ischemic Heart Failure Extension Study (STICHES), which showed a statistically significant difference in survival in favor of the CABG group at 11 years (59% vs. 66%; HR: 0.84; 95% CI: 0.73–0.97; p = 0.02) (4).
However, in STICH, there was a high rate of treatment crossover between the CABG and GDMT groups, which potentially could have diluted overall treatment efficacy and statistical power. In particular, of the 602 patients assigned to medical therapy, 17% underwent CABG and another 6% received PCI. Most of these crossovers occurred immediately post randomization following the results of the viability testing and may have attenuated the benefits of surgical revascularization in earlier time periods (e.g., within 1 year). Therefore, the aim of this study was to re-analyze the STICH data and evaluate the “as-treated (AT)” effect of CABG versus GDMT alone using two statistical methods to correct for confounding by selective treatment crossovers without excluding randomized patients and to compare these analyses to those published. This analysis was prepared using STICH Research Materials obtained from the NHLBI Biologic Specimen and Data Repository Information Coordinating Center and does not necessarily reflect the opinions or views of the STICH or NHLBI.
The methods we used include time-dependent Cox proportional hazards modeling, and instrumental variable (IV) analysis with G-estimation, both of which accounted for crossovers within the first year post-randomization. In the time-dependent Cox regression, we estimated the HR of CABG vs GDMT adjusted for strata of randomization and baseline risk factors as prespecified in the original STICH analyses. We varied a landmark time point through 1-year to account for the immortal time bias. For the instrumental analysis with G-estimation (5), we fitted one logistic regression model for randomization group, while regressing on randomization stratum, and one Cox proportional hazards model for the outcome of interest, while regressing on randomization stratum, randomization group and receipt of CABG. Statistical analyses were performed with R version 4.0.3 (R Foundation for Statistical Computing, http://www.r-project.org/).
Hazard ratios for both analyses for the primary endpoint (all-cause mortality at 6 years after randomization) were compared with those originally reported for the intention-to-treat (ITT), AT and per-protocol (PP) analyses. We additionally estimated HRs for secondary endpoints of cardiovascular mortality and readmissions. In the original AT analysis, regardless of initial randomization, patients treated with GDMT throughout the first year were categorized as the GDMT group, whereas patients who underwent CABG during the first year were categorized as CABG. The original PP analysis defined the GDMT group as patients randomly assigned to GDMT who did not crossover to CABG during the first year of follow-up, and the CABG group as patients randomly assigned to CABG who actually underwent CABG. The effect sizes estimated from these analyses could have suffered from immortal time bias because group membership was determined late after randomization without allowing for delayed entry.
Our analysis included a total of 1212 patients. The published ITT analysis included 610 patients randomized to CABG plus GDMT and 602 to GDMT alone. The published AT analysis included 592 without CABG and 620 patients with CABG in the first year (HR 0.70; 95% CI 0.58–0.84; P<0.001). The published PP analysis included 537 patients without CABG and 555 patients with CABG in the first year (HR 0.76; 95% CI 0.62–0.92; P = 0.005).
At 6 years of follow-up, CABG plus GDMT resulted in significantly lower risk of all-cause death in the time-dependent Cox regression (HR: 0.72; 95% CI: 0.60–0.88) and IV analyses (HR: 0.72; 95% CI: 0.54–.96). Results were overall consistent using different time cutoffs through 1-year to define treatment crossovers (Figure 1). CABG was also associated with significantly lower risk of cardiovascular mortality, and the combined endpoint of any death or cardiovascular rehospitalizations, irrespective of the method to account for treatment crossover.
Figure 1.
Treatment effect of CABG vs. GDMT alone based on time-dependent Cox regression and IV approaches with varying cut-off time point for receipt of CABG
Hazard ratios (solid lines) and 95% confidence intervals (shaded area) are shown for receipt of CABG based on the multivariable Cox regression analysis including a time-dependent covariable (top) and instrumental variable analysis (bottom). The analyses were repeated by varying the time period of CABG receipt post-randomization from time of randomization through 1-year.
In conclusion, in this re-analysis of the STICH trial in which we accounted for treatment crossovers, coronary revascularization using CABG surgery within the first year after randomization was associated with improved outcomes, including all-cause mortality, that already emerged at 6 years of follow-up. These findings suggest, that among surgically-eligible patients with coronary anatomy amenable to revascularization, patients with severe ischemic cardiomyopathy should be considered for CABG at any time in order to derive the potential benefits from surgical revascularization.
Acknowledgments
Funding:
This study was supported by cooperative agreement (U01 HL088942) funded by the National Heart Lung and Blood Institute, the National Institutes of Neurological Disorders and Stroke of the National Institutes of Health (NIH), Bethesda, MD, and the Canadian Institutes for Health Research (CIHR), Ottawa, ON, Canada. The views expressed in this article are those of the authors and do not necessarily represent the views of the National Heart, Lung, and Blood Institute; National Institutes of Health; or the United States Department of Health and Human Services.
Role of Funder/Sponsor:
The NIH was instrumental in providing funding to obtain the data set and the support of the data coordinating center which preformed the data analysis, interpretation, preparation and review of the manuscript.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.Patterson T, McConkey HZR, Ahmed-Jushuf F, Moschonas K, Nguyen H, Karamasis GV, et al. Long-Term Outcomes Following Heart Team Revascularization Recommendations in Complex Coronary Artery Disease. J Am Heart Assoc. 2019;8(8):e011279. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Yancy CW, Jessup M, Bozkurt B, Butler J, Casey DE Jr., Drazner MH, et al. 2013 ACCF/AHA guideline for the management of heart failure: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol. 2013;62(16):e147–239. [DOI] [PubMed] [Google Scholar]
- 3.Velazquez EJ, Lee KL, Deja MA, Jain A, Sopko G, Marchenko A, et al. Coronary-artery bypass surgery in patients with left ventricular dysfunction. N Engl J Med. 2011;364(17):1607–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Velazquez EJ, Lee KL, Jones RH, Al-Khalidi HR, Hill JA, Panza JA, et al. Coronary-Artery Bypass Surgery in Patients with Ischemic Cardiomyopathy. N Engl J Med. 2016;374(16):1511–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Sjolander A, Martinussen T. Instrumental Variable Estimation with the R Package ivtools. Epidemiologic Methods. 2019;8(1). [Google Scholar]